Testing procedures for the P91 steel

The work on the P91 steel can be divided into two stages. The first stage of the experimental work involved the investigation of the P91 parent metal. Most of the effort on this material was carried out in order to get a better understanding of the behaviour of this martensitic type material in cyclic loading conditions at high temperatures. Several types of tests were carried out on the P91 specimens, including isothermal cyclic plasticity, thermomechanical fatigue and notched-bar testing.

_____________ Experimental work 3.4.1 Isothermal cyclic plasticity test of P91 steel

Strain-controlled testing was used to investigate the behaviour of P91 under cyclic plasticity conditions in isothermal conditions. The aims of the tests are firstly, to obtain the stress-strain behaviour of the material so that a constitutive material behaviour model can be developed based on these data. Secondly, a fatigue life model, particularly under low cycle fatigue conditions, can be developed. The solid TMF specimens (i.e. Fig. 3.2) were used in these tests.

The test loading conditions were based on the high temperature requirement of power plant and to produce a deformation which will give plastic behaviour. The isothermal tests of P91 specimens were conducted at 400, 500 and 600°C. At each temperature, displacement controlled tests with a strain amplitude of ±0.5% and a strain rate of 0.1%/s were performed in fully reversed tension-compression conditions. The strain amplitude value was selected to ensure that significant plasticity would occur with this amplitude. Using the same strain amplitude at each temperature, tests with a hold period, in tension, for 2 minutes were performed in order to study the time-dependency effect. The duration of hold period was chosen so that constant rate of stress drop, in each cycle, was recorded in the test. In addition, several strain-controlled tests, with the same strain rate, were performed at 600°C with strain amplitudes of ±0.2%, 0.25% and 0.4%.

Actions performed prior to starting the test

Prior to starting up each test, temperature calibration of the TMF machine system was performed in order to obtain a temperature variation within the gauge length of

±10°C of the target temperature. This can be achieved by designing the copper coil for the P91 steel specimen by adjusting its shape particularly the number of coil turns and the pitch between each coil turn. The characterization of the copper coil

shape controls the density of magnetic flux, induced by the induction heating system, and thus determines the specimen’s temperature. Five thermocouples were welded on the gauge section, as shown in Figure 3.5, to measure the temperature during thermal calibration. A coil from previous tests on a RR1000 specimen, as shown in Figure 3.6(a) which gave temperature gradient of within ±10°C for the RR1000, was used as a starting design shape. However, that coil design produced a temperature variation of more than ±15°C, even after adjusting the pitch of coil turn. Another shape of coil as shown in Figure 3.6(b) was used and there was an improvement for temperature reading. The final design of the coil shape for P91 specimen tests is shown in Figure 3.7; this gave the required temperature variation along the gauge section as plotted in Figure 3.8.

For the P91 steel isothermal strain-controlled tests, the temperatures during the tests were measured by using thermocouples (TC). Two K-type thermocouples were used in each test; one thermocouple acted as the main TC which controls the required temperature for a test and another thermocouple was used to check the reading of the main TC before starting the test. The thermocouple wires were attached to the specimen’s surface by spot welding and the distance between the two thermocouple wires must be less than 1mm as suggested by Hahner et al.

(2008). If the TC readings indicate an unacceptable variation of temperature, the qualities of the spot welds are checked and repetition of the spot welding process may be performed if necessary.

From previous tests on RR1000, using the TMF machine system (Hyde, 2010), cracks were found at the TC spot weld positions on the gauge section. In order to avoid that situation, temperature in the strain-controlled tests of P91 steel was measured by positioning thermocouples away from the gauge section, i.e., on the

_____________ Experimental work shoulder of the specimen. Thus, the temperature ratio between the centre of the gauge section and the specimen’s shoulder were recorded, at the position shown in Figure 3.9(a). Average temperature readings at the shoulder position for corresponding gauge section temperatures of 400, 500 and 600°C, obtained from the setup of this experiment, were 344, 429 and 516°C, respectively. For the actual tests, both thermocouples were placed on the specimen’s shoulder, as shown in Figure 3.9(b), and the temperature ratios were used in the test.

In order to obtain a good response for a specified test loading, the PID (proportional, integral, derivative) values of the TMF machine system were tuned. The strain amplitudes and testing periods for the actual tests were applied on a specimen and the PID values were adjusted accordingly to get a good fit to controlled shape parameter (strain, load or temperature). The values of P 36, I 2 and D 0 were used for the P91 steel tests determined by trial-and-error process. This procedure was performed using the temperature calibration specimen.

Actual test

Once the temperature calibration and PID tuning procedures were completed, the actual tests were performed, using new P91 specimens. Thermocouples were instrumented on the specimen as shown in Figure 3.9(b) and then, the specimen was installed on the test rig, as shown in Figure 3.7. A verification of Young’s modulus was performed at room temperature to ensure correct operation of the extensometer. The applied load in the Young’s modulus verification was small, to ensure that it was within the elastic limit, in order to avoid an irreversible displacement in the specimen before starting the test. Correct operation of the extensometer resulted in a reasonable smooth plot of linear data and the measured Young’s modulus values were comparable to reference values at the same

temperature. After verifying the extensometer set-up, the specimen was heated up to the required temperature.

The specimen was heated up to the required temperature within 30 seconds from room temperature. The specimen was continued to be heated without applying any load for up to 5 minutes in order to get stabilized temperatures throughout the solid specimen. During that time, the mechanical strain waveform was set as required and a failure criterion, which set the condition for the machine to stop, was defined as a given percent decrease of maximum stress from certain cycle. From the results of the first test on a P91 specimen at 600°C with ±0.5% strain amplitude, using a 50% stress decrease from the 50^th cycle as the failure criterion, the specimen were found to totally fracture and the extensometer fell down. This may cause damage of

the extensometer if it occurs regularly. Thus, a lower percent of stress decrease, i.e., 30% stress drop from the 50^th cycle, was defined as the failure criterion for the

remaining tests and the tests were found to stop without causing total fracture of the specimen. Visible cracks were observed on all specimens’ surfaces. Test data were recorded automatically by the Instron TMF software.

Figure 3.5: The positions of the 5 thermocouples along the gauge section during temperature calibration stage

_____________ Experimental work

Figure 3.6: Different designs of coil which were used in the intial stage of temperature calibration, (a) a coil used in previous RR1000 tests and (b) a new coil

made and tested for P91 test specimens140

Figure 3.7: TMF test machine set-up used for P91 specimens with finalised design of the induction heating coil and the extensometry

589 593 597 601 605

480 500 520 540 560

Time (seconds)

Temperature (C)

Eurotherm TC1 TC2 TC3 TC4 TC5

Figure 3.8: Temperature reading at 600°C (normal) during the temperature calibration stage from the 5 thermocouples on the gauge section where TC1 to TC5

represent the locations of the thermocouples from top to bottom

Figure 3.9: The positions of thermocouples on the gauge section and the shoulder of the specimen for (a) measuring the temperature ratio and (b) during the

strain-controlled tests of the P91 steel

_____________ Experimental work

3.4.2 Thermo-mechanical fatigue testing of P91 steel

The aim of the thermomechanical fatigue (TMF) testing of the P91 steel specimens was to observe the effects of cyclic mechanical loading and temperature on the behaviour of the steel. The test data obtained from the TMF tests were also used to verify the viscoplasticity model developed in this study for predicting anisothermal behaviour. The tubular specimen, as shown in Figure 3.3, was used for the TMF tests. A mechanical strain range of ±0.5% was used for all of the TMF tests for two temperature ranges, i.e., of 400-500˚C and 400-600˚C. The TMF tests were carried out under in-phase loading conditions (maximum temperature at maximum strain) up to failure and out-of-phase loading conditions (minimum temperature at maximum strain), as shown schematically in Figure 3.10. All of the TMF tests were carried out with a period of 60s per cycle.

0 15 30 45 60

Time (seconds) Mechanical Strain, m

-mechanical T: in-phase T: out-of-phase

and Temperature, T (°C)

Figure 3.10: Schematic representation of anisothermal strain-controlled loading with in-phase and out-of-phase temperature cycles

An infra-red pyrometer was used instead of thermocouples at the specimen shoulders to measure temperatures in the TMF tests. The controlled temperature at the specimen’s shoulder causes difficulty in determining the temperature along the gauge section when temperature is cycled. A pyrometer is a non-contacting device for measuring the temperature and hence surface crack due to spot-welding can be avoided. However, an oxide layer, which forms during testing, may affect the accuracy of temperature measurement (Beck and Rau, 2008). An initial study was implemented to assess the severity of the oxide layer formed during the test by preoxidising the specimen at 650°C for 24 hours. It was found that the P91 specimen surface was not significantly oxidised compared to other specimens used in previous tests in the lab such as stainless steel and RR1000 specimens. Thus, the effect of oxidation on pyrometer reading was neglected in this study.

In order to use the pyrometer in the test, the emissivity values for the required temperature between 400 and 600°C were determined prior to starting the TMF test.

For this purpose, the initial thermal cycling study was performed by attaching 5 thermocouples on the gauge section’s surface to measure the temperature variation, as shown in Figure 3.5. The temperatures of the specimens in the tests were controlled by the use of the middle thermocouple. The specimen was heated to a specified temperature and the temperature was kept constant for several minutes to stabilize the temperature of the specimen. The corresponding emissivity value was then recorded. This process was repeated at other temperatures within the required temperature range. The recorded emissivity values at 400, 500 and 600°C are 0.615, 0.625 and 0.635 respectively. Each emissivity value was then used in a temperature cycling test, controlled by pyrometer. This was done in order to get an emissivity value which can give a better temperature reading for the specified temperature range and to ensure that the maximum temperature

_____________ Experimental work variations within gauge length of the specimen are within the required range, i.e.,

±10°C (Hahner et al., 2008). The emissivity value of 0.625 was used in the TMF tests.

For the thermomechanical fatigue test, the TMF software divides the test into four stages, namely, stabilisation, measurement, verification and test stage. The first three stages were the preparatory stages in which only a temperature waveform was applied and resultant thermal strains were measured and verified. The thermal strain was measured by using the extensometers data and recorded at the base time interval. In the test stage, the mechanical strain cycle of ±0.5% was defined in the TMF software; the mechanical strain can only be realized by controlling the total strain applied to the specimen considering the thermal strain measured in the preparatory stages. For example, the total strain applied to a specimen in the tension direction, for TMF in-phase tests was higher than the defined mechanical strain due to the contribution of thermal strain (expansion of specimen due to increase of temperature). In contrast, the thermal and mechanical strains were opposite in direction for the TMF out-of-phase tests and hence resulted in shorter applied total strain in the test. The 30% stress drop from the 50^th cycle was also used as the failure criterion for the TMF test.

3.4.2 Cyclic notched bar test of P91 steel

The cyclic notched bar tests on P91 steel were performed using a fully reversed load-controlled condition in order to verify the capability of the material model under multiaxial conditions. The nominal stress amplitude was 300MPa on the minimum notch section and a constant temperature of 600°C was applied to the specimens throughout the test. The stress value was chosen in order to produce a similar strain range effect, on the notch section, as that in the isothermal strain-controlled tests on

a P91 solid specimen at 600°C. The TMF notched specimen, as shown in Figure 3.4, was used for the tests. Two types of loading were used for the tests: firstly, the load was continuously applied in triangular waveform with a cyclic period of 20 seconds; and secondly, a dwell period of 2 minutes, which is the same duration as the cyclic test on the P91 solid specimen, was introduced in the previous triangular waveform at peak loading. Each test was performed until failure occurred.

For the notched bar tests, thermocouples were used to measure the test specimen temperature. However, the position of the main TC was not on the shoulder of specimen. The main TC was attached on the gauge section which was 2 mm above the notch corner as shown in Figure 3.11. During thermal calibration, two additional thermocouples were attached on the centre of the notched surface. The ratio between the main TC and the thermocouples reading for notch section was recorded, making sure that the temperature was 600°C at the centre. In the actual tests, the main and guide TC were attached to new P91 notched specimens without any TC on notched radius.

Figure 3.11: 4 thermocouples were attached to the gauge section during thermal calibration

_____________ Experimental work